Skip to content

High confidence phishing email in Office 365, and what to do with it

On this page

A high confidence phishing email in Office 365 is one that Microsoft 365’s filter gave its strongest phishing verdict. It means the filter is close to certain the message is a phishing attempt. The message goes to quarantine instead of anyone’s inbox, and by default only an admin can see it or let it out. Allow lists, Outlook Safe Senders and mail flow rules can’t override it.

So if you’ve met the phrase, the message is held, and no one at the firm has it in front of them. That’s the system doing its job. What’s left is a short set of decisions:

  • Is it real phishing? Then leave it where it is. It deletes itself on a timer.
  • Is it a real message from a client or supplier? Then an admin releases it and tells Microsoft it was wrong, in one step.
  • Does your team need to know when this happens? That’s a quarantine setting, and the default is silence.

The rest of this page covers each, with Microsoft’s own wording, checked on October 4, 2026.

What a high confidence phishing email means in Office 365

Microsoft’s filter sorts incoming mail into five verdicts, listed on its anti-spam protection page:

  • Spam
  • High confidence spam
  • Phishing
  • High confidence phishing
  • Bulk

Phishing and high confidence phishing are the same suspicion at two strengths. Plain phishing usually lands in Junk Email or in quarantine, depending on your policy. High confidence phishing has one destination. Microsoft puts it like this:

As part of secure by default, high confidence phishing messages are always quarantined. Users can’t release their own quarantined high confidence phishing messages, regardless of any available settings configured by admins.

In practice, the message never reaches Outlook. It sits in quarantine in the Microsoft Defender portal at security.microsoft.com, under Email & collaboration > Review > Quarantine. Its Quarantine reason reads “High confidence phishing.” If you’re new to quarantine itself, what a quarantine folder is covers where it lives and why you can’t search for it from Outlook.

Microsoft doesn’t publish the exact signals behind the verdict. What it does document is one cause you control. A block entry in your own Tenant Allow/Block List turns any matching message into high confidence phishing. That comes up again in the runbook below.

Can you get phished just by opening an email?

A high confidence phish can’t reach you that way, because it never arrives. It’s held before delivery.

Sometimes the filter only becomes sure after a message has landed. For that, Microsoft runs zero-hour auto purge (ZAP). The ZAP page says that for “read or unread messages that are identified as high confidence phishing after delivery, ZAP quarantines the message.” It looks back over the last 48 hours of delivered mail. It also doesn’t tell the user, so a message can quietly disappear from someone’s inbox. If a colleague says an email “vanished,” check quarantine first.

For ordinary mail that does arrive, reading it in a current version of Outlook is rarely where the harm happens. Harm comes from what you do next: pressing a link, opening a file, or answering. How to spot a phishing email walks through those checks in order.

How high confidence phishing relates to the spam confidence level (SCL)

The spam confidence level is a number from -1 to 9. Microsoft’s filter stamps it into a message header called X-Forefront-Antispam-Report. For years it was the main way to read a verdict. Admins still meet it in older guides and in mail flow rules.

Microsoft’s SCL page now says this about cloud mailboxes:

The value doesn’t determine whether spam filtering identifies a message as Spam or High confidence spam, and it doesn’t determine the action taken on the message.

So the SCL won’t tell you whether a message was judged high confidence phishing. Here’s what each value is still used for:

SCLWhat it means in Microsoft 365 today
-1A mail flow rule asked to bypass most spam filtering
0 or 1The message was evaluated and not treated as spam
5 or 6A mail flow rule can set these to treat a message as spam
9A mail flow rule can set this to treat a message as high confidence spam

Microsoft says a value of 5 or higher “generally indicates the message is considered bad.” It also says SCL’s main job now is supporting on-premises Exchange, including hybrid setups.

What SCL -1 means

SCL -1 is the bypass value. A mail flow rule set it, usually to allowlist a sender or to accept mail that another filter already checked. The header shows SFV:SKN when a rule bypassed spam filtering.

Here’s the one misreading that costs firms real mail. SCL -1 doesn’t make a message safe from the high confidence phishing verdict. The anti-spam page says it in five words, about rules that bypass filtering: “High confidence phishing messages are still filtered.” A rule can wave a sender past spam checks. It can’t wave them past this.

Reading the verdict from the header

To see what really happened to a message, read the CAT value in the same header, per Microsoft’s anti-spam message headers page:

Header valueWhat it tells you
CAT:HPHSH or CAT:HPHISHHigh confidence phishing
CAT:PHSHPhishing
CAT:HSPMHigh confidence spam
CAT:SPMSpam
CAT:SPOOFSpoofing
SFV:SKNA mail flow rule bypassed spam filtering
SFV:SKQThe message was released from quarantine

Pasting a raw header into Microsoft’s Message Header Analyzer lays these out in a readable list. The Authentication-results line next to them shows whether SPF, DKIM and DMARC passed. If the message claims to come from Microsoft itself, the genuine Microsoft senders, side by side lists the real domains next to the common fakes.

Why your allow list can’t override it

Microsoft calls the policy “secure by default.” Its secure by default page says some of your firm’s overrides “aren’t applied for malware or high confidence phishing.” Those are:

  • Allowed sender and allowed domain lists in anti-spam policies
  • Outlook Safe Senders, including the ones your staff added themselves
  • The IP Allow List in the default connection filter policy
  • Exchange mail flow rules, also known as transport rules

The same page is plain that this isn’t optional:

Secure by default isn’t a setting that you can turn on or off.

Microsoft gives its reason too. Its data shows a user is “30 times more likely to click a malicious link in messages in the Junk Email folder versus Quarantine.” That’s why the old Move message to Junk Email folder action no longer applies to this verdict. If an anti-spam policy still has it selected, the message is quarantined anyway.

If a sender you trust keeps landing here, adding them to Safe Senders or writing an SCL -1 rule won’t change anything. The fix is a submission, covered below.

The exceptions Microsoft supports

Microsoft lists four situations where overrides are appropriate. Each has its own tool, and none of them is a mail flow rule.

SituationWhat Microsoft says to useFits a small firm?
A phishing simulation from a non-Microsoft training vendorThe advanced delivery policy, listing the vendor’s sending domains, IPs and URLsYes, if you run awareness tests
A security mailbox that must receive raw, unfiltered mailA SecOps mailbox in the same advanced delivery policyRarely; this is for security teams
Another filter sits in front of Microsoft 365Secure by default only applies when your MX record points to Microsoft 365Only if you’ve chosen a separate gateway
A real message caught by mistakeAdmin submission, with Allow this messageYes, this is the common case

The third row needs care. When your domain’s MX record points to a non-Microsoft service, rules and allow lists can deliver high confidence phish to inboxes. That moves the whole job of catching it onto the other filter. It’s a design choice about your mail path, and not a workaround for one stuck sender.

Who can release it: the quarantine settings

What a user can do with a quarantined message is set by a quarantine policy. Each anti-spam verdict has one attached. For high confidence phishing, the default is a policy named AdminOnlyAccessPolicy. Microsoft’s quarantine policies page describes its effect:

SettingAdminOnlyAccessPolicy (the default)A custom policy with notifications and request release
Can the recipient see the message in quarantine?NoYes
Does the recipient get a notification?NoYes, within 4 hours, daily or weekly
Can the recipient release it?NoNo, they can only request release
Who decides?An adminAn admin approves or denies the request

That third row is fixed. Even a policy that grants full release rights is downgraded for this verdict. Microsoft’s wording:

If the policy is configured for users to release these quarantined messages, users are instead allowed to request the release of these quarantined messages.

The default has a cost for a small firm. A client’s real invoice can sit in quarantine and expire without anyone at the firm knowing it existed. A custom quarantine policy with notifications on and request release allowed keeps the admin as the gate. It also tells the recipient that something is waiting. For most small firms that’s the better balance. Assign it to the high confidence phishing verdict in your anti-spam policy.

How long a held message lasts

Quarantine has a clock. Per Microsoft’s quarantine retention table, messages quarantined by anti-spam policies, including high confidence phishing, are kept for:

  • 15 days in the default anti-spam policy and in policies you create yourself. You can set this anywhere from 1 to 30 days.
  • 30 days in the Standard and Strict preset security policies, and you can’t change it there.

After that, Microsoft says, “the messages are permanently deleted and can’t be recovered.” If nobody’s notified, 15 days goes quickly. That’s a reason to set retention to 30.

When a real message is quarantined as high confidence phish

This is the job most admins arrive with. A client says they sent something, and it’s sitting in quarantine with the scariest label Microsoft has. Here’s the order that keeps you safe and gets the mail moving.

  1. Confirm it with the sender first. Call them on a number you already had, not one from the email. A high confidence verdict is usually right, and a message from a client’s stolen account looks exactly like the client.
  2. Find it in quarantine. In the Defender portal, go to Email & collaboration > Review > Quarantine, then filter Quarantine reason to high confidence phishing. Check the Expires date.
  3. Read the header. Look for CAT:HPHSH, then at the Authentication-results line. Failed SPF, DKIM or DMARC points to the sender’s setup, and the fix starts on their side.
  4. Check your own block list. Microsoft’s Tenant Allow/Block List page warns that a block entry for a URL or domain can mark a message as high confidence phishing “even if the sender is legitimate.” If someone blocked a link-shortener or a file-sharing domain last year, this is often why. Remove the entry and the problem stops at the source.
  5. Release it and report it in one step. Select the message, then Release email. Tick Submit the message to Microsoft to improve detection (false positive), then Allow this message. That last option adds allow entries for the sender and any URLs or files in the message.
  6. Choose how long the allow lasts. The default is 45 days after last used date. It renews each time the filter meets that sender and finds them clean. You can pick 1, 7 or 30 days, or a date up to 30 days away.
  7. Confirm it arrived. The entry takes effect within 5 minutes. Use message trace to confirm the released message reached the inbox. It shows up with the release time, not the original send time, so tell the recipient to look at the top of the inbox.
  8. Release the others by hand. Microsoft’s false positive guide says it plainly: “Quarantined messages aren’t released automatically.” Search quarantine for the same sender and release each one.

Steps 5 and 6 are the only supported allow for this verdict. The Tenant Allow/Block List page says “you can’t create allow entries directly” for high confidence phishing. The submission is the route. Your role needs to be Security Administrator or equivalent to see the submit and allow options.

A few days after release, check the Submissions page for Microsoft’s verdict on what you reported. If it found a cause in your own configuration, fix that, and the allow entry stops mattering.

When the sender’s domain is new

New domains get caught more often, both a client’s and your own. Microsoft doesn’t publish the reasons. The fixes it does document all point the same way: authenticate the domain properly, and report the false positive so the filter learns.

A client or supplier on a new domain

A firm that has just rebranded, or a new supplier on a domain registered last month, has no sending history. Microsoft’s false positive guide says to “ask the external sender to fix their SPF, DKIM, and DMARC records to align with their sending infrastructure.” Send them that request, release what’s waiting using the steps above, and keep the allow entry on the 45-day default while their records settle.

If their invoices come from an accounting or e-signature platform, that platform has to be included in the sender’s SPF and signing with their domain’s DKIM. That’s usually the missing piece.

A domain your own firm just added

When you add a domain to Microsoft 365 and start sending from it, publish SPF, DKIM and DMARC for it before the first message goes out. Do the same for any service that sends as your domain, like a newsletter tool or a booking system. Mail that claims your domain but fails authentication looks like spoofing, and Microsoft’s headers flag that case with a composite authentication reason of 601 or 010.

Avoid the tempting shortcut of adding your own domain to an allowed domains list. Microsoft’s anti-spam page says those lists bypass SPF, DKIM and DMARC checks, and “never add common domains” applies just as much to your own. An attacker can spoof it as easily as anyone.

When it really is phishing

Most of the time, the verdict is right. Then the job is small:

  • Leave it in quarantine. It deletes itself when the retention period ends.
  • Deny any release request with a short note to the colleague about why.
  • Check who else it was sent to. The quarantine list shows every recipient, which tells you whether the firm was targeted as a group.

If someone at the firm was involved in the same campaign by another route, such as a Teams message, a phone call or a link they did click, what to do if you’ve already clicked covers the next hour.

Where Point sits

Point works with the mail Microsoft 365 delivers. A message quarantined as high confidence phish never reaches the mailbox, so it never reaches Point either. That part stays with Microsoft and with whoever administers your tenant.

For the mail that does arrive, Point sets risky and malicious messages aside before they reach your feed. You look at them when you’re ready. The release decision for anything in Microsoft’s quarantine is still the admin’s, through the steps above. What else Point handles for a small firm is on the benefits page.

Common questions

What does “high confidence phishing” mean?

It’s Microsoft 365’s strongest phishing verdict. The filter is close to certain the message is phishing, so it quarantines it. Only an admin can release it, and allow lists, Safe Senders and mail flow rules can’t override that.

Can I get phished just by opening an email?

Not by a high confidence phish, because it’s held before it reaches you. For mail that does arrive, the risk sits in what you do next: a link, a file or a reply.

What does the spam confidence level (SCL) mean in Office 365?

It’s a number from -1 to 9 that Microsoft’s filter stamps in the message header. In cloud mailboxes Microsoft says it no longer decides the verdict or the action. Read the CAT value in the same header instead.

What does it mean when the SCL is set to -1 in Microsoft 365?

A mail flow rule asked Microsoft to bypass most spam filtering for that message. It doesn’t bypass the high confidence phishing verdict, so a message with SCL -1 can still be quarantined.

Can a user release a high confidence phishing message?

No. Even if the quarantine policy allows release, users can only request it, and an admin approves or denies the request.

How long does Microsoft 365 keep a quarantined high confidence phish?

Fifteen days by default, adjustable from 1 to 30 in your anti-spam policy. In the Standard and Strict preset policies it’s 30 days. After that it’s permanently deleted.

Ready for a calmer inbox?

Join the private beta

We're onboarding a few teams at a time. Leave your email, confirm it once, and we'll send an invitation the moment a place opens.

By joining you agree to our privacy policy.

Private beta

What you're joining

It runs on the mail you have

Point sits on top of Gmail or Outlook. Your address, your history and your contacts stay exactly as they are, so there is nothing to migrate.

You set how much Point does

Out of the box everything waits for your review, replies included. You hand over only what you trust, one kind of work at a time.

Join the private beta

We're onboarding a few teams at a time. Leave your email, confirm it once, and we'll send an invitation the moment a place opens.

By joining you agree to our privacy policy.